2Abstract The 52 nd round of National Sample Survey data was used to explore gender differences in the living arrangements of older persons in India. Data demonstrate that older women, aged 60 and older are more likely to live alone than older men, after controlling for demographics, socioeconomic characteristics, health status, economic independence, and property ownership. Economic independence is a statistically significant determinant of living arrangements for both older men and women. Worse health in terms of physical immobility increases the likelihood of living alone or with spouse only for older men. Results indicate that having sons decreases the likelihood of living alone for older women, particularly if they are immobile. In contrast, immobility has no significant association with living arrangement choice for older men who have no children or only have daughters, and instead significantly increases the likelihood of living alone for older women with only daughters.
3Introduction Populations worldwide, with the exception of the African continent, are rapidly growing older as a result of declining fertility rates, improvements in life expectancy, and medical breakthroughs. Populations in developing countries that have recently experienced demographic transitions are projected to age faster than those in developed nations (Palloni, 2000), which will challenge the developing world to grapple with economic development and aging concurrently. The lessons learned from developed countries might not be able to guide policymakers, because the nature of challenges in the developing world is different. Hence, studying the various facets of aging that are unique to developing countries is critical to help inform policy makers. The Madrid International Plan of Action on Aging of 2002 (United Nations, 2002) pointed to the topic of living arrangements of older persons as one of the areas requiring major thrust because living arrangements of older persons is a key aspect of their well being. Co residence might be viewed as a fundamental means of ensuring daily needs of the older population in the absence of social security nets as well as a source of mutual support for younger and older adults in certain cultures. Knowledge is limited regarding the actual living conditions and how they affect the well being of older populations in different countries. The key questions that we explore in this paper are: what are the determinants of living arrangements for older Indians? Are there any gender differences in living arrangements for older persons in India? To what extent do economic independence and other socioeconomic factors influence living arrangement choice? Early literature regarding this topic encompasses sociological and anthropological works that theorize about living arrangement choice (Goode, 1963). In the last few decades, empirical studies
4attempted to find correlates of living arrangement choice for older persons in different parts of the world. More recently, the United Nations published two reports (United Nations, 2001 and 2005) that provide an overview of living arrangements of older persons worldwide as well as research agendas that can help fill the knowledge gap in this area. Although studies are available on the topic of older persons' living arrangements in Korea (Lee and Palloni, 1992; Martin, 1989), Taiwan (Hermalin, 2002; Hermalin and Yang, 2004), Singapore, Indonesia, (Frankenberg et al., 2002), Malaysia (DaVanzo and Chan, 1994; Martin, 1989), Fiji, Philippines (Martin, 1989), Vietnam (Friedman et al., 2003), Lesotho (Mba, 2002), Latin America (DeVos and Holden, 1988), sub Saharan Africa (Zimmer and Dayton, 2005), i South Asia (Martin, 1990) rural Pakistan (Kochar, 2002) , and rural Bangladesh (Ghuman and Ofstedal, 2004), very few articles are available that specifically study the living arrangements of older Indians. India has the second largest and the fastest growing older population in the world. The decadal growth rate of this group between 1991 and 2001 was almost double that of the general population (Sharma and Xenos, 1991, JaiPrakash, 1999). According to the Indian Census projections ii , number of older persons in India was 71 million in 2001 and is projected to rise to 173 million by 2026. This will increase the proportion of aged Indians from 6.9% in 2001 to 12.4% in 2026. Hence, what France has experienced during the past 100 years in terms of an aging population, India will have to grapple with in 25 years. Unlike the West, India is still steeped in the traditional ways of living where multiple generations reside in the same household. Older and retired Indians expect to live with their children as a source of security. With improved health infrastructure and increased life expectancy, older persons are expected to live longer. At the same time that the population of India is steadily increasing, immense changes in standards of living and levels of development are occurring. Joint family systems are on the decline and more urban families are becoming nuclear. More rural families are relenting to the pressures of a developing society with increased rural to urban migration of younger working adults. Longer life
5spans will create pressures on families to fund postretirement requirements of their parents and grandparents. Although India has had a long tradition of providing pension support, this has been limited exclusively to those who ever worked in the organized sector, which covers only 11% of the older population (Chanana and Talwar, 1987). Certain Indian states have policies regarding providing older indigent persons financial support, but this system is under funded and the older population is grossly underserved (Chanana and Talwar, 1987; Ponnuswami, 1999; Dandekar, 1996). An important facet of aging is the gender composition among older persons. Women tend to marry older men; hence, women are more likely to be widowed and live in the widowed state for longer periods. The earning capacity of women, and consequently, their ability to maintain themselves, is less than that of men. In India, unlike other countries, sex ratio favors males for persons 60 and older (Visaria, 2001), which reveals incidence of higher mortality and lower status of women. However, more women than men live to be age 80 years or older. As a result, the majority of older Indian women live longer than their spouses, live as widows, live in poverty, and frequently live as dependents of their children or extended families. Given this background, exploring the nature of living arrangements among older Indians, with a special focus on gender differentials appears to be an important task. Previous research on older Indians had been limited because of a lack of availability of large scale data that could provide enough details on the older persons. Recent studies have examined the extent of well being of older persons (Cain, 1986; Vlassof, 1990; Dandekar, 1996; Ponnuswami, 1999; Chanana and Talwar, 1987, Rajan, 2006), provided a description of living arrangements of the aged (Rajan and Kumar, 2003) and analyzed the connection between living arrangements and poverty among older rural residents (Pal, 2003; Pal, 2006). However, to the best of our knowledge no previous studies exist that examine the gender differences in living arrangements among older adults in India.
6This paper addresses this gap by analyzing the socioeconomic correlates of the living arrangement choices of older persons in urban and rural India, focusing in particular on gender differences. Data from the 52 nd round of National Sample Survey (NSS) are used for this analysis. In this paper, older persons have been defined as those who are aged 60 years or older, which is consistent with the definition of older persons used in other parts of the world (Visaria, 2001, Palloni, 2000). Although most studies have analyzed the likelihood of older co residence with children and others as a source of old age support, we take a slightly different approach in this paper by seeking to determine the statistically significant correlates that might explain the likelihood of older persons living alone. We define 'living alone' as those who live alone or with a spouse only. This kind of living arrangement is widespread in developed countries and on the increase in the developing world. In India, approximately one in seven older Indians belonged to this group according to the 52 nd NSS data (See Table 1). They are also likely to be the most vulnerable in the face of an income or health shock or even a disaster. In this paper, we analyze who these older persons are and what are the factors associated with living alone. This might be an important exercise in helping inform policymakers about certain groups at high risk. The analysis is timely because India is engaged in careful reassessment, planning, and implementation of national retirement policies. Careful attention to the needs of the growing proportions of aged, especially older widows is rapidly becoming the need of the day. Methodology Theoretical Underpinnings Different theories have offered explanations for determinants of living arrangements. In the absence of government supported social security schemes or private pension plans, traditional living arrangements, where older persons live with their children and extended families, provide a form of
7social safety net. Classical theories purported that multigenerational shared living arrangements became less common with industrialization, largely as a result of a loss of status and control by older persons (Cowgill and Holmes, 1972). However, these classical theories did not withstand empirical testing, which led to alternative explanations (Elman and Uhlenberg, 1995). The main explanations offered were demographic, economic, and cultural. Demographically, changes in mortality, fertility, migration, and marriage can explain living arrangements. Although kin availability and kin composition can influence living arrangement choice toward co residence (Elman and Uhlenberg, iii 1995; Palloni, 2000), an increase in life expectancy and reduction in fertility that leads to more older persons and fewer children to look after them, can lead to decreases in co residence (Wolf, 1984). Urbanization and migration can also reduce likelihood of co residence (Mba, 2002). Desire for reciprocity, mutual exchange, economies of scale, and greater security from economic risks are suggested economic explanations that determine living arrangement choice. With modernization and ability to earn higher incomes, the trend is toward dissolution of traditional living arrangements. Privacy tends to become a normal good, and economically independent older persons tend to prefer living alone, not with the extended family (Costa, 1997). In contrast, high cost of living can encourage co residence even after controlling for income (DaVanzo and Chan, 1994). Studies have reported that after controlling for socioeconomic characteristics, number of married children decreases the likelihood of living alone; homeownership and higher income increases the likelihood of living alone; disability decreases the likelihood of living alone; and education increases the likelihood of living alone (Martin 1989, Palloni, 2000). Traditional obligations and duties, sense of family values, and cultural factors are non economic reasons that can explain living arrangement choice. Certain cultures expect filial piety and sons to take responsibility for older parents, which might explain greater likelihood of co residence when sons are present. A study from the Indian state of Maharashtra demonstrated that older widows are happiest when
8living with their sons, while regularly receiving visits from their daughters (Vlassof, 1990). A study of older Hispanic persons in the United States indicates that cultural desirability factors are most important in determining living in institutional facilities (Burr and Mutchler, 1992). These existing theories and findings are useful as they offer testable hypotheses regarding living arrangement choice for older persons in the Indian context. We applied economic modeling to explain the different living arrangement choices of older persons. In addition to covariates that are determined from the existing theories, we use covariates such as economic independence and property ownership to proxy for financial and emotional empowerment for the aged. Bargaining models of the family would suggest that older persons and adult children can negotiate a living arrangement to optimize their objectives. Economically independent older persons or those with property probably negotiate a transfer to their children in return for free goods and services that they might otherwise have to buy. This means that economically independent older persons are also more socially and emotionally involved and likely to co reside with their children and extended family. However, this can mean that their living space is more constrained with less privacy. If privacy were a normal good, one might predict that economically independent older persons would choose to live on their own. How economic independence affects living arrangements is therefore an empirical question to explore. Altruism models predict that household income matters and not individual incomes. Children might live with their parents not because of financial reasons but because of the knowledge that parents are taken care of. Alternatively, children might send transfers of income to support their parents. Economic independence for older persons might be a substitute for this income transfer, in which case, older persons might choose not co reside. In contrast, economic independence or transfers might have
9no influence on co residential choice if some unobservable reason drives co residence. Hence, influence of economic independence on living arrangements is again an empirical question. Further, studies demonstrate that family composition matter more than family size (Wolf and Soldo, 1988), and if more than one child is present, bargaining between siblings can determine a parent's living arrangements and lead to a breakdown in co residence (Pezzin et al., 2007). Depending on the cultural expectation, whether children are sons or daughters will influence living arrangement outcomes. Hence, number and sex composition of children are critical determinants of living arrangements, driven empirically. Data Data were taken from the 52 nd round of the National Sample Survey, conducted by the National Sample Survey Organization (NSSO) of India, which primarily focuses on health care and education, including the problems of older persons (60 years). The survey period was July 1995 June 1996, and data were collected from all Indian states and union territories, except Andaman and Nicobar Island, Dadra and Nagar Haveli, Lakshadweep, and certain remote areas of Arunachal Pradesh and Nagaland. A stratified two stage random sampling design was adopted. The first stage units (FSU) were villages (determined by the National Census of 1991) in the rural sector and blocks (depending on population, as determined by NSSO) in the urban sector. The second stage units are households in both sectors. Of each FSU, 10 households were selected independently from the second stage grouping by circular systematic sampling with a random start. The survey provides information on all the older members of the households, including number of sons and daughters, number of dependents, economic status and activities during work life, living arrangements, whether physically immobile, disabilities, self reported health status, chronic health
10conditions, reasons for retirement, retirement benefits, ownership and management of assets, management of social matters, religious matters, and daily chores and availability of food, clothes, and medicines. The sample consists of 26,643 households. Total number of older persons is 34,084 (5.4 percent), of which 17,211 (50.5 percent) were men and 16,873 were women. Empirical Estimation Our empirical model is a reduced form logistic regression that estimates the likelihood of living alone for men and women separately. We used a dummy dependent variable that equals 1 if living alone, and 0 otherwise. An older person is defined as 'living alone' if the person is living with a spouse and no other kin or is unmarried and living with no other kin (in an old age home or not). Palloni (2000) uses the same definition in a survey of living arrangements of older persons. When living with at least one child (or other kin), 'co residence' or 'not living alone' is used. The NSSO question on living arrangement choice captures information on whether the older person lives alone, in an old age home or not, whether he or she lives with spouse only, whether he or she lives with spouse and children, whether he or she lives with children only, or whether he or she lives with other relatives or nonrelatives. The explanatory variables used were age (age of the person in years), age2 (age squared), female (equals 1 if female; 0 if male), curmar (equals 1 if currently married; 0 otherwise ), scstcode (equals 1 if person belongs to scheduled caste or scheduled tribe; 0 otherwise), urban (equals 1 if urban; 0 if rural), kids ( total number of sons and daughters), nolit (equals 1 if not literate; 0 otherwise iv ), quart2 (those in the second quartile of household per capita consumption expenditure), quart3 (those in the third quartile), quart4 (those in the fourth quartile), reference category if in the lowest socioeconomic quartile, ownprop (equals 1 if the person owns property, either managing or not managing that property, and 0 if the person does not own